meok-mcp-injection-scan-mcp
meok-mcp-injection-scan-mcp
2026年4月のCVE波で公開されたプロンプトインジェクション / ツールポイズニング / SSRFクラスの脆弱性について、あらゆるMCPサーバーをスキャンします。
pip install meok-mcp-injection-scan-mcpなぜこれが存在するのか
2026年4月はMCPにとって悪い月でした。Anthropicは、約7,000の公開サーバー(約1億5,000万ダウンロード)に影響を与える「設計上の」MCP RCEクラスを公開しました。mcp-server-gitはCVEチェーンを出荷しました。DockerDashはインジェクションチェーンによって侵害されました。ツール記述のプロンプトインジェクション(「ツールポイズニング」)は、すべての主要なMCPホストに対して実証されました。
MCPサーバーを本番環境で運用している場合、あるいは導入前に監査を行う場合、2026年4月の開示で対象となったパターンをフラグ立てする高速なスキャンが必要です。このMCPがそのスキャンを行います。
Related MCP server: agent-audit
チェック内容
5段階の深刻度にわたる30以上の標準ルール:
CRITICAL — 直接的なRCE、システムプロンプトのオーバーライド、認証情報の流出パターン、デフォルト設定内のシェルメタ文字、
file:/// 内部ネットワークURL(DockerDashの169.254.169.254メタデータピボットベクトル)。HIGH — エンコードされたペイロード、エージェントへの命令的指示、サプライチェーンプロンプト、環境変数参照、ツールシャドウイング。
MEDIUM — 緊急性 / 権威的な言語、
additionalProperties=true、無制限の文字列、ツール名のなりすまし。LOW — 長すぎる説明、ゼロ幅 / 双方向オーバーライド文字(U+202E PoCベクトル)。
カバレッジは、OWASP LLM Top 10、GenAI Red Team v1、2026年4月のAnthropic MCP RCE開示、およびmcp-server-git CVEチェーンに対応しています。
公開されているツール
ツール | 目的 |
| リモートMCPサーバーのツールリストを取得してスキャンする |
| 貼り付けられたJSONツールリストをスキャンする(認証が必要なサーバー用) |
| 調達グレードの署名付き証明書を発行する(Proティア) |
| サブスクライブ前にルールカタログ全体を検査する |
| サブスクリプションリンク + ティア比較 |
料金
ティア | 価格 | 内容 |
Free | £0 | 1日5スキャン、署名付きレポートなし |
Starter | 無制限スキャン + 署名付きレポート | |
Pro | + 定期的な再スキャン + 48時間サポート | |
Enterprise | + カスタムルールパック + 4時間SLA |
すべての署名付き証明書は https://meok-attestation-api.vercel.app/verify/<cert_id> に存在し、監査人や調達チームはアカウントなしで確認できます。
得られないもの
これは静的パターン・スキャナーです。動的な汚染分析を実行したり、敵対的な入力でサーバーをファジングしたり、人間のレッドチームに取って代わるものではありません。これは監査の最初の80%を、5秒で、無料で行うものです。
MEOK AI Labsによる構築
ソロ創業者。ロンドン。PyPIで234のMCPパッケージ。ライブ署名インフラストラクチャは meok-attestation-api.vercel.app。ストアフロントは councilof.ai。カタログの取得: https://meok-attestation-api.vercel.app/catalogue。
流通チャネル
PyPI:
pip install meok-mcp-injection-scan-mcp(本パッケージ)Apify Store (イベントごとの支払い): https://apify.com/knowing_yucca/meok-mcp-injection-scan
GitHub (ソース): https://github.com/CSOAI-ORG/MEOK-LABS/tree/main/mcps/meok-mcp-injection-scan-mcp
Available Tools
5 toolsaudit_tool_descriptionsA
Audit a JSON string containing a tool list (paste from your own MCP server's tools/list output). Same rule catalogue as scan_mcp_url — useful when the server is behind auth or not yet deployed.
tools_json accepts either: a raw list, or {"tools": [...]}, or
{"result": {"tools": [...]}}.
Behavior: This tool is read-only and stateless — it produces analysis output without modifying any external systems, databases, or files. Safe to call repeatedly with identical inputs (idempotent). Free tier: 10/day rate limit. Pro tier: unlimited. No authentication required for basic usage.
When to use: Use this tool for security assessment, threat detection, or vulnerability analysis. Suitable for automated security scanning and risk evaluation.
When NOT to use: Do not rely solely on this tool for production security decisions. Always combine with manual security review.
Args: tools_json (str): The tools json to analyze or process. api_key (str): The api key to analyze or process.
Inputs: tools_json (required), api_key.
list_rulesA
List every detection rule in the canonical catalogue. Useful for buyers auditing what we check before subscribing.
Behavior: This tool is read-only and stateless — it produces analysis output without modifying any external systems, databases, or files. Safe to call repeatedly with identical inputs (idempotent). Free tier: 10/day rate limit. Pro tier: unlimited. No authentication required for basic usage.
When to use: Use this tool for security assessment, threat detection, or vulnerability analysis. Suitable for automated security scanning and risk evaluation.
When NOT to use: Do not rely solely on this tool for production security decisions. Always combine with manual security review. Behavioral Transparency: - Side Effects: This tool is read-only and produces no side effects. It does not modify any external state, databases, or files. All output is computed in-memory and returned directly to the caller. - Authentication: No authentication required for basic usage. Pro/Enterprise tiers require a valid MEOK API key passed via the MEOK_API_KEY environment variable. - Rate Limits: Free tier: 10 calls/day. Pro tier: unlimited. Rate limit headers are included in responses (X-RateLimit-Remaining, X-RateLimit-Reset). - Error Handling: Returns structured error objects with 'error' key on failure. Never raises unhandled exceptions. Invalid inputs return descriptive validation errors. - Idempotency: Fully idempotent — calling with the same inputs always produces the same output. Safe to retry on timeout or transient failure. - Data Privacy: No input data is stored, logged, or transmitted to external services. All processing happens locally within the MCP server process.
Inputs: no named parameters.
pricingB
Pricing + subscribe links for this MCP.
Behavior: This tool is read-only and stateless — it produces analysis output without modifying any external systems, databases, or files. Safe to call repeatedly with identical inputs (idempotent). Free tier: 10/day rate limit. Pro tier: unlimited. No authentication required for basic usage.
When to use: Use this tool for security assessment, threat detection, or vulnerability analysis. Suitable for automated security scanning and risk evaluation.
When NOT to use: Do not rely solely on this tool for production security decisions. Always combine with manual security review. Behavioral Transparency: - Side Effects: This tool is read-only and produces no side effects. It does not modify any external state, databases, or files. All output is computed in-memory and returned directly to the caller. - Authentication: No authentication required for basic usage. Pro/Enterprise tiers require a valid MEOK API key passed via the MEOK_API_KEY environment variable. - Rate Limits: Free tier: 10 calls/day. Pro tier: unlimited. Rate limit headers are included in responses (X-RateLimit-Remaining, X-RateLimit-Reset). - Error Handling: Returns structured error objects with 'error' key on failure. Never raises unhandled exceptions. Invalid inputs return descriptive validation errors. - Idempotency: Fully idempotent — calling with the same inputs always produces the same output. Safe to retry on timeout or transient failure. - Data Privacy: No input data is stored, logged, or transmitted to external services. All processing happens locally within the MCP server process.
Inputs: no named parameters.
scan_mcp_urlA
Fetch a remote MCP server's /tools/list (or any JSON tool listing) and scan every tool's name + description + inputSchema against the canonical 30+ injection-pattern rules.
Returns a structured report: per-tool findings, severity counts, score 0-100, verdict, and remediation hints. Free tier: 5 scans/day per key.
Behavior: This tool is read-only and stateless — it produces analysis output without modifying any external systems, databases, or files. Safe to call repeatedly with identical inputs (idempotent). Free tier: 10/day rate limit. Pro tier: unlimited. No authentication required for basic usage.
When to use: Use this tool for security assessment, threat detection, or vulnerability analysis. Suitable for automated security scanning and risk evaluation.
When NOT to use: Do not rely solely on this tool for production security decisions. Always combine with manual security review.
Args: url (str): The url to analyze or process. api_key (str): The api key to analyze or process.
Behavioral Transparency: - Side Effects: This tool is read-only and produces no side effects. It does not modify any external state, databases, or files. All output is computed in-memory and returned directly to the caller. - Authentication: No authentication required for basic usage. Pro/Enterprise tiers require a valid MEOK API key passed via the MEOK_API_KEY environment variable. - Rate Limits: Free tier: 10 calls/day. Pro tier: unlimited. Rate limit headers are included in responses (X-RateLimit-Remaining, X-RateLimit-Reset). - Error Handling: Returns structured error objects with 'error' key on failure. Never raises unhandled exceptions. Invalid inputs return descriptive validation errors. - Idempotency: Fully idempotent — calling with the same inputs always produces the same output. Safe to retry on timeout or transient failure. - Data Privacy: No input data is stored, logged, or transmitted to external services. All processing happens locally within the MCP server process.
Inputs: url (required), api_key.
signed_safety_reportA
Issue a cryptographically signed safety report for the scanned MCP server. Returns a cert with a public verify URL anyone can hit to confirm the audit happened on the date claimed.
Pro / Enterprise tier only.
Behavior: This tool is read-only and stateless — it produces analysis output without modifying any external systems, databases, or files. Safe to call repeatedly with identical inputs (idempotent). Free tier: 10/day rate limit. Pro tier: unlimited. No authentication required for basic usage.
When to use: Use this tool for security assessment, threat detection, or vulnerability analysis. Suitable for automated security scanning and risk evaluation.
When NOT to use: Do not rely solely on this tool for production security decisions. Always combine with manual security review.
Args: subject (str): The subject to analyze or process. findings_json (str): The findings json to analyze or process. score (int): The score to analyze or process. note (str): The note to analyze or process. api_key (str): The api key to analyze or process.
Behavioral Transparency: - Side Effects: This tool is read-only and produces no side effects. It does not modify any external state, databases, or files. All output is computed in-memory and returned directly to the caller. - Authentication: No authentication required for basic usage. Pro/Enterprise tiers require a valid MEOK API key passed via the MEOK_API_KEY environment variable. - Rate Limits: Free tier: 10 calls/day. Pro tier: unlimited. Rate limit headers are included in responses (X-RateLimit-Remaining, X-RateLimit-Reset). - Error Handling: Returns structured error objects with 'error' key on failure. Never raises unhandled exceptions. Invalid inputs return descriptive validation errors. - Idempotency: Fully idempotent — calling with the same inputs always produces the same output. Safe to retry on timeout or transient failure. - Data Privacy: No input data is stored, logged, or transmitted to external services. All processing happens locally within the MCP server process.
Inputs: subject (required), findings_json, score, note, +1 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Each tool targets a distinct aspect of the security scanning workflow: 'audit_tool_descriptions' handles local JSON analysis, 'scan_mcp_url' handles remote scanning, 'list_rules' lists detection rules, 'pricing' provides subscription details, and 'signed_safety_report' generates a signed certificate. No two tools have overlapping purposes.
The naming convention is mostly snake_case but inconsistent in structure: three tools follow a verb_noun pattern ('audit_tool_descriptions', 'list_rules', 'scan_mcp_url'), while 'pricing' is a bare noun and 'signed_safety_report' uses a past participle adjective-noun form. This mix reduces predictability.
With 5 tools, the server is well-scoped for its purpose of scanning MCP servers for injection vulnerabilities. Each tool serves a clear function without overlap, and the count is appropriate for a focused security assessment tool.
The tool surface covers the core workflow: local audit, remote scan, rule listing, pricing, and signed reporting. Minor gaps exist (e.g., no tool for configuration or result history), but these are not essential for the primary use case. Overall, the surface is sufficiently complete for its domain.
Maintenance
Related MCP Connectors
Security scanner for MCP servers. Detect vulnerabilities, prompt injection, and tool poisoning.
Security scanner for AI agents and MCP servers: injection, jailbreak, tool abuse.
Audit your MCP servers for poisoned tools and leaked secrets, and check AI-generated code for flaws.
Scan any MCP server for tool-poisoning, security, auth & license. Trust score before install.
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceScans MCP servers for prompt injection, supply chain attacks, excessive permissions, and code execution risks. Includes an offline blacklist that catches known-compromised packages like LiteLLM 1.82.7/1.82.8 and Trivy with zero latency.19MIT
- AlicenseNot gradedqualityCmaintenanceSecurity scanner for MCP servers. Detects prompt injection, command injection, auth bypass, and excessive permissions across tools, resources, and prompts.45 npm3MIT
- AlicenseNot gradedqualityCmaintenanceScans MCP servers, AI agent skills, and plugins for 68+ malicious patterns including credential exfiltration, prompt injection, and code execution.5MIT
- AlicenseBqualityDmaintenanceAutomated security red-team for any MCP server that scans manifests against OWASP LLM Top 10 and MCP-specific risks, returning a 0-100 hardening score and HMAC-signed report.7MIT